Accommodating multiple potential normalizations in microbiome associations studies

BMC Bioinformatics

Table 4 Estimated power of the tests under Setting 1, where \(\delta\) is the mixing proportion

		None	Rarefaction	TSS	CSS	CLR	Omnibus
\(\delta =0\%\)	Linear Regression	0.943	0.842	0.844	0.991	0.999	0.996
	ZINQ	0.882	0.738	0.902	0.882	0.992	0.974
	QRank	0.780	0.821	0.799	0.809	0.943	0.900
\(\delta =10\%\)	Linear Regression	0.738	0.613	0.616	0.924	0.976	0.948
	ZINQ	0.658	0.493	0.689	0.651	0.923	0.833
	QRank	0.526	0.576	0.551	0.564	0.761	0.571
\(\delta =20\%\)	Linear Regression	0.415	0.332	0.336	0.658	0.793	0.690
	ZINQ	0.348	0.241	0.371	0.346	0.650	0.497
	QRank	0.245	0.276	0.261	0.267	0.417	0.328
\(\delta =30\%\)	Linear Regression	0.148	0.120	0.122	0.268	0.364	0.271
	ZINQ	0.114	0.084	0.125	0.117	0.265	0.164
	QRank	0.086	0.098	0.090	0.092	0.140	0.108

ISSN: 1471-2105